ABSTRACT 1. INTRODUCTION

Transcription

1 Modeling, optimization, and experimental validation of a resonant piezo-optical ring sensor for enhanced active and passive structural health monitoring Erik Frankforter, Jingjing Bao, Bin Lin, Victor Giurgiutiu Department of Mechanical Engineering, University of South Carolina Columbia, SC 29208, frankfor@ .sc.edu, baoj@cec.sc.edu, linbin@cec.sc.edu, victorg@sc.edu ABSTRACT A mechanical resonant piezo-optical ring sensor was studied, designed to selectively enhance the response of piezoelectric wafer active sensors (PWAS) and fiber Bragg grating (FBG) sensors. The frequency characteristics of the ring sensor were modeled through modal and harmonic analyses. The models were used to guide the experimentation, serving as a basis for comparison and implementation. Pitch-catch, resonance, and acoustic emission (AE) experiments were performed to compare the performance of the ring sensor to plate-mounted PWAS and FBG. Factors relating to optimal in-service implementation, particularly symmetric placement of FBG and PWAS, were investigated. It was found that the ring sensor was capable of amplifying an incoming Lamb wave signal. This was applied to AE experiments, where selective frequencies were amplified such that the time-domain signal had a larger amplitude response. Keywords: SHM, PWAS, FBG, AE, Acoustic Emission, Resonance, Guided Wave, Sensor 1. INTRODUCTION Structural health monitoring (SHM) implements structurally-embedded smart sensors to detect and characterize damage 1,2. SHM sensors are permanently installed onto a structure, having the potential for continuous in-service assessment of a structural state. This is in contrast to the approach taken by nondestructive evaluation (NDE) which typically requires structures be taken offline for manual scanning at periodic intervals. SHM allows for a continually updated damage state, which can reduce service costs over the lifecycle of a structure. SHM also has the potential to enable a high level of confidence regarding structural integrity, even to the point that SHM sensors may offer useful life predictions beyond that of more conservative life span designs. SHM techniques can be broadly classified into two approaches - active and passive. In active SHM, a transducer generates a waveform which interacts with damage and is sensed by a receiver (e.g. pitch-catch, pulse-echo, and phased array techniques). In passive SHM, a sensor listens for signs of damage which propagate through the structure such as impact and acoustic emission (AE) generated waves. A number of sensors have shown great promise as SHM sensors; one such sensor is the piezoelectric wafer active sensor (PWAS) which can act in both active and passive modes. PWAS have the ability to send and receive ultrasonic waves when bonded to a surface through electromechanical interaction from the piezoelectric effect. PWAS are well-suited for SHM due to their reliability, minimally invasive profile, and low cost. Another promising SHM sensor is the fiber Bragg grating (FBG) sensor, which can serve as a receiver in active SHM or for passive SHM 3,4. A FBG sensor embedded into an optical fiber uses a periodic grating in its refractive index to create a wavelength-dependent filter. Upon strain of the longitudinal axis of the fiber, the filtered wavelength shifts; these shifts can be tracked and related back to strain. Much of the technical challenge in SHM is not only in detecting a waveform, but in subsequent analysis and interpretation. The gold standard in SHM is analysis is in basing interpretation on the physics of a waveform. For thin plates, such as those found in aerospace structures, this involves the analysis of Lamb waves which propagate through the structure. The multi-modal nature of Lamb waves makes this a nontrivial problem, as each mode propagates differently through a structure, and these modes must be accounted for, separated, or eliminated.

2 In this paper, we assess a sensor designed to tackle this issue. The sensor is a combination of FBG, PWAS, and an auxiliary structure tuned to confer advanced sensing functionalities. Through development of a sensor with advanced sensing functionalities, we hope to simplify an incoming Lamb wave such that the characteristics of the incoming signal may be well-known for straightforward analysis. 2. PIEZO-OPTICAL RING SENSOR In this research, we assess a resonance amplifying piezo-optical ring sensor. The concept of the piezo-optical ring sensor has been disclosed and investigated in previous works 5,6, and in a US patent application 7. The sensor has been dubbed a piezo-optical ring sensor, or ring sensor for short; the ring referring to its shape, which may work in concert with mounted piezoelectric and FBG sensors. The ring was developed through model-driven design, focusing on mechanical resonance amplification principles. Through its geometry and sensing mechanisms, this gives it advantages over surface-mounted FBG sensors: Omnidirectionality the capability to sense guided waves from all directions Mode selectivity the capability to sense vertical wave motion and reject in-plane motion Frequency tuneability the capability to select for frequencies where an incoming signal is predominant, and reject other frequencies Two ring sensor designs have been prototyped, having fundamental frequencies at 100 khz and 300 khz. These are shown in Figure 1. Figure 1: Piezo-optical ring sensors. The ring sensors depicted have resonances near 100 khz (left) and 300 khz (right) The ring itself may be seen as a tuned auxiliary structure, to which PWAS and FBG are bonded. The PWAS is bonded to the flat surface on the top of the ring and the FBG is stretched and bonded across a hole symmetrically placed through the ring. This work offers a number of improvements to the ring sensor both as a manufactured sensor and as a design concept. We have advanced the finite element modeling (FEM) of the ring sensor, particularly of its mode shapes and frequency response. We have also optimized the ring sensor eliminating confounding effects from bonding the ring to the plate and bonding the sensors to the ring. Finally, we have shown the capability of the ring to mechanically amplify a guided wave, both during pitch-catch experiments and from acoustic emission events. 3. FINITE ELEMENT MODELING OF THE RING SENSOR Experimental studies have shown the ring sensor has significant advantages in terms sensing functionalities over plate-mounted FBG sensor or PWAS. To this end, we investigated the dynamic response of the ring sensor in ANSYS Workbench Two types of analysis were performed on free aluminum and stainless steel 100 khz ring sensors: 1. harmonic analysis, to represent the frequency response of the ring sensor to a harmonic excitation, and 2. modal analysis, to represent the natural frequencies and mode shapes of the ring sensor.

3 Amplitude, m Amplitude, m A damping ratio of 0.01% was used across all modes. A total of elements and nodes were used, which was shown to be large enough to represent the highest frequency mode shape in the frequency range assessed. The ring sensor was modeled under two boundary conditions: one fully free, and another with the displacement of the bottom flat surface fully constrained. This was to test the effect of bonding the ring sensor. 3.1 Harmonic analysis of the ring sensor The harmonic analysis assessed the response of the ring sensor to excitation on the flat surface. Two harmonic 1 N line forces were placed on one of the flat surfaces, at the edges between the flat and curved surfaces of the ring. The forces were directed parallel to the hole drilled for the FBG, and opposite in direction. This simulates a pin-force model for a PWAS located on top of a ring sensor, or potentially a base excitation for a ring sensor bonded to the plate. The displacement component along the FBG hole was measured at the holes drilled for the FBG on each side. The responses for each side were subtracted to get a differential displacement, which is related to what would be sensed by the FBG sensor. The harmonic analysis results were almost identical for the stainless steel and aluminum ring sensors, with the only variation being in the amplitude. This is because the two materials have nearly the same wave speed. The harmonic response of the aluminum ring sensor under both free and fixed-bottom conditions is shown in Figure 2. x Fundamental frequency, 107 khz Second harmonic, 273 khz x Fundamental frequency, 72 khz Second harmonic, 210 khz a) Frequency, khz b) Frequency, khz Figure 2: Harmonic response of the aluminum ring sensor under (a) fully free conditions and (b) with the bottom flat surface displacement fully constrained The free ring sensor shows a sharp resonance at 107 khz with a quiescent range as low as 0 khz and up to the second resonance at 273 khz. The ring sensor with the bottom displacement fixed showed a drop in the fundamental resonance frequency to 72 khz with a moderate drop in amplitude. The frequency of the higher harmonics also dropped to approximately 210 khz, and the amplitude of the higher harmonics increased to over half the fundamental resonance. This indicates that beyond just modeling the free ring sensor, the dynamic response of the ring sensor after bonding should be taken into account during future design. 3.2 Modal analysis of the ring sensor A modal analysis was performed for the free ring sensor to characterize the harmonic response. Of particular interest was whether modes could be eliminated through judicious sensor placement onto the ring structure. A few representative displacement mode shapes for the 100-kHz aluminum ring sensor can be seen in Figure 3. From Figure 3a,c the first two resonances identified in the harmonic response have a breathing type motion, where the resonance expands and contracts along the axis of the fiber. Figure 3b, which shows a mode not represented in the harmonic response, has vanishing displacements at the FBG hole and at the top of the ring where the PWAS is located. This is due to two factors: the symmetry of the ring and the placement of the sensors along axes of symmetry. A number of modes within the quiescent range of the ring sensor can be characterized by a similar effect. This can be seen in the experimental studies in the next section.

4 a) b) c) Figure 3: Mode shapes of the free aluminum ring sensor at (a) 107 khz, (b) 122 khz, and (c) 273 khz 4. EXPERIMENTAL ANALYSIS OF THE FREE RING SENSOR Under ideal circumstances, the frequency spectrum of the ring sensor would follow the designed spectrum. However, a number of practical difficulties presented themselves in terms of implementation. These are: bonding of the PWAS and FBG to eliminate extraneous modes, and the effect of bonding the ring sensor to the structure. 4.1 Equipment setup PWAS excitations in the form of Hanning windowed tone bursts and chirp excitations were generated using an HP33120A function generator. The tone bursts were particularly used for pitch-catch experiments across a plate, where the chirp excitations were used to assess the frequency response of the ring sensor. The chirp signals consisted of a harmonic excitation with a frequency varying linearly with time. The time-frequency relationship is depicted in Figure 4. Since the frequency increases linearly with time, a time-domain signal is directly related to its frequency components, where frequency components can be identified directly from the time-domain response. Figure 4: Time-frequency relation of a linear chirp excitation. The chirp signal has a harmonic excitation with its frequency increasing linearly over time For strain sensing with the FBG, we follow previous work of Lin 8 using an intensity modulation approach, where a tunable laser system (Luna Phoenix 1400 TLS) in conjunction with an optical circulator was used to scan the wavelength spectrum of the FBG, and then fix on the half-maximum of the falling FBG peak. From the intensity modulation, as discussed by Norman 9, as the wavelength shifts, this is related to the intensity of the signal through the FBG spectrum s falling slope. By dividing the received voltage by the slope, and with parameters from the FBG and optical equipment, the strain can be directly calculated. This intensity was relayed by a New Focus 2053 photodetector and sent to a Tektronix TDS5034B oscilloscope, which also served as a signal digitizer for the PWAS signals. 4.2 Bonding the FBG to the ring sensor As noted in the harmonic analysis of Figure 2, there were a number of mode shapes proximal to the desired mode at 100 khz; however, these modes could be eliminated by exploiting symmetry of the ring sensor. An experiment was performed where a chirp signal was sent from a PWAS bonded to the top of a ring sensor and received by FBG on the ring sensor. The PWAS was overhanging for ease of connecting the ground wire (Figure 5a). When sensed by an FBG on the side of the ring sensor, the overhanging PWAS was able to excite multiple modes, and the FBG could sense these modes (Figure 5b). When the same excitation was sensed by an FBG bonded through the central symmetric hole (Figure

5 5c), the peak near 100-kHz dominates, where additional modes are sensed below this frequency, albeit at a very low amplitude. a) b) c) Figure 5: Effect of mounting FBG to the ring sensor. (a) A free ring sensor has a side-mounted FBG and a centrally-mounted FBG, with the PWAS overhanging for grounding purposes. (b) Response of the side-mounted FBG to a linear chirp excitation from khz. (c) Response of the centrally-mounted FBG to a linear chirp excitation from khz 4.3 Bonding the PWAS to the ring sensor To eliminate the need for an overhanging PWAS, a PWAS was bonded symmetrically to the top of the aluminum ring sensor (Figure 6a), with the grounding done via conductive epoxy on the ring sensor. A linear chirp excitation was sent from the PWAS and received by the centrally placed FBG fiber. From the response (Figure 6b), the extraneous modes seen from the overhanging PWAS were eliminated through the combination of symmetric transducer and receiver placement. The single-mode response was considered ideal, and this specimen was used for the rest of the experiments. a) b) Figure 6: Aluminum ring sensor with optimized sensor placement. (a) PWAS and FBG are bonded symmetrically to an aluminum ring sensor ground by conductive epoxy. (b) FFT of the linear chirp response, sent from the PWAS to the FBG 4.4 Effect of bonding the ring sensor to a plate To test the effect of bonding the aluminum ring sensor to a structure, the response to a linear chirp excitation was assessed. The chirp was transmitted from a PWAS bonded to the top of the ring sensor and received by the FBG through the central hole of the ring sensor. Under free boundary conditions, the 100-kHz aluminum ring sensor showed a close match to the FEM harmonic response, with a single sharp peak at 107 khz (Figure 7a). When mounting this same aluminum ring to a plate, the spectrum spreads out, with a fundamental resonance dropping to 84 khz and multiple resonances appearing in the vicinity (Figure 7b). The drop of the resonance frequency can be explained by the harmonic analysis in Figure 2. By bonding the ring sensor to a plate, the boundary condition is closer to the fixed boundary condition in Figure 2b. Thus, a stiffer material for a ring sensor may be more desirable in future redesigns. At minimum, the effect of bonding the sensor should be taken into account during model-driven design.

6 a) b) Figure 7: Change in frequency spectrum upon bonding a ring sensor to a plate. (a) A free ring sensor shows a sharp, dominant resonance at 107 khz. (b) Upon bonding to a plate, the resonance to approximately 84 khz, with the appearance of additional modes 5 EXPERIMENTAL ANALYSIS OF A PLATE-MOUNTED RING SENSOR In the previous sections, we assessed sensor dynamics through wave propagation experiments across the ring sensor. In this section, we assess the sensor under conditions closer to the sensor implementation, testing the response to Lamb waves propagating across a thin plate for pitch-catch, resonance, and acoustic emission experiments. 5.1 Experimental setup To determine the response of the ring sensor to a Lamb wave for SHM applications, it was bonded to a 1200 mm x 900 mm x 1.2mm 2024-T3 aluminum plate. The ring sensor was bonded to the plate using a Vishay Micro- Measurements M-Bond 200 adhesive kit. The experimental setup can be seen in Figure 8a,b. In Figure 8a, two transmitter PWAS are bonded 15cm away from a sensor cluster, one with a propagation path longitudinal to the fiber axes, and one transverse to the fiber axes. The sensor cluster has four sensors: 1) a PWAS on the plate, 2) an FBG on the plate, 3) an aluminum 100-kHz ring sensor outfitted with PWAS and FBG, 4) a stainless steel 100-kHz ring sensor outfitted with PWAS and FBG. As shown in Figure 7b, wave-absorbing clay (McMaster # 6102T21) was placed around area of interest to reduce or eliminate edge reflections. A 9 cm wide layer was spread onto the plate on top and bottom. Using this technique, only one edge reflection was seen, with amplitude 96% lower than the initial incident wave. This was considered sufficient to significantly reduce the effect of standing waves across the plate for longer-duration excitations, making the predominant effect measured that of the Lamb waves. a) b) Figure 8: Experimental setup of aluminum plate. (a) PWAS longitudinal and transverse to the FBG fiber are used to excite the sensor packet. (b) Wave-absorbing clay is used to eliminate or greatly reduce the effect of edge reflections 5.2 Pitch-catch experiments A series of pitch-catch experiments were performed across the aluminum plate specimen to assess the ring sensor response to an incoming transient signal. The aluminum ring sensor, which was outfitted with the symmetrically placed

7 FBG and PWAS, was compared to the PWAS on the plate. A 20V peak-to-peak 3-count Hanning windowed tone burst from the longitudinal transmitter PWAS was used as an excitation. A tuning curve the variation of amplitude with frequency of the A0 Lamb wave mode was assessed for a frequency range of khz. The response of the PWAS on the plate follows the tuning characteristics of PWAS transducer and receiver (Figure 9a). In comparison, the ring sensor tuning dominates, with a clear single peak near its designed frequency for both PWAS and FBG (Figure 9b,c). As in the chirp excitation experiments across the ring sensor, the resonance frequency is lowered to approximately 82 khz due to the effect of bonding to the plate. Of particular note is that the tuning curves from pitch-catch experiments across the plate are particularly smoother in nature than the chirp excitation experiments across the ring sensor. This is explained by the fact that a Hanning windowed tone burst has a spread in its frequency content about its center frequency, incorporating dynamics effects of the ring sensor in a neighborhood around its center frequency. a) b) c) Figure 9: Tuning curves for 3-count Hanning windowed tone bursts sent from the longitudinal transmitter PWAS. The maximum value of the response s envelope was used as the amplitude. (a) Tuning curve for plate-mounted PWAS receiver. (b) Tuning curve for the aluminum ring-mounted PWAS receiver. (c) Tuning curve for the aluminum ring-mounted FBG receiver In Figure 10, waveforms received from the FBG on the ring sensor are shown, excited by the longitudinal and transverse PWAS as 75 khz 3-count Hanning windowed tone bursts. Even though the excitation is off-resonance, the ring sensor is still able to detect the wave. This is due to both the spread of the plate-mounted ring sensor frequency response, and the spread of the frequency contents of the Hanning windowed excitation about its center frequency. The response to the transverse excitation is still rather large in magnitude, even though the FBG sensor only sensed strain along the axis of the fiber. Additionally, a significant amount of ringing can be seen after the arrival of the A0 wave packet. This indicates that energy is being stored within the ring. This indicates that the resonance principles that the ring was designed from may be able to take effect given a sufficient length excitation. a) b) Figure 10: 75kHz 3-count Hanning windowed pitch-catch signals sent by longitudinal and transverse transmitter PWAS to the FBG on the aluminum ring sensor. (a) Excitation by longitudinal PWAS. (b) Excitation by transverse PWAS

8 5.3 Resonance experiments To assess the capability of the ring sensor to amplify an incoming signal, we assessed experiments where we could gradually approach a continuous harmonic wave. The longitudinal transmitter PWAS was used to transmit a wave across the plate to the FBG on the plate and the FBG on the ring sensor. By changing the number of counts of a Hanning windowed tone burst, the frequency spread and length of an excitation could be controlled. As the number of counts increased, the frequency spread about the center frequency narrowed and the signal length increased. For these experiments, the wave-absorbing clay was critical, as if standing waves developed across the plate, the sensor response would depend on its location (e.g. at a local maximum or a node). Ideally, the amplitude of the incoming waveform should asymptotically approach that of the continuous sine wave. Since the exact excitation frequency was critical due to the narrowing frequency spread, a continuous sine was used to find the frequency which produced a maximum response on the ring sensor, found to be 82.1 khz. The variation in receiver strain can be seen in Figure 11. The voltage received was divided by the FBG falling slope so it is directly comparable to strain, and normalized by the response of the continuous sine wave from the FBG on the ring sensor. The FBG on the plate had a larger strain for low-count tone bursts. After a certain number of counts, the strain detected by the FBG on the ring sensor surpasses the strain from the FBG on the plate. There are a number of ways to interpret this. One possibility is that the frequency content of the signal needed to fall sufficiently close to the resonance frequency of the ring sensor for a large enough amplification. Another possibility is that the additional boundary created by bonding the ring sensor to the plate added additional attenuation to the wave, which needed to be overcome by stored energy from a longer duration signal. This is the first case where the ring sensor was shown to be able to amplify the response of an incoming wave. A capacity for amplification is a positive indicator for the design approach. This indicates that future redesigns and optimizations of the ring sensor, currently being assessed, have a potential for greater amplification based on these principles. Figure 11: Resonance results for plate-mounted FBG and ring sensor-mounted FBG. The strain was normalized to the strain from the ring sensor-mounted FBG

9 5.4 Acoustic emission experiments After the demonstration of resonance capabilities of the ring sensor, a natural follow-up was whether this effect could be used for passive SHM, particularly with AE signals which have frequency contents in a range which can be amplified by the ring sensor. To this effect, pencil lead break (PLB) simulated AE signals were generated, using 0.5mm HB lead, on the aluminum plate in Figure 7 at distances 10cm longitudinal and transverse to the ring sensor. The response of the FBG on the plate and FBG on the ring sensor are shown in Figure 12. The case of a PLB exciting the FBG on the plate from a transverse direction is omitted, as the FBG could not sense a significant component of the wave. a) b) c) d) e) f) Figure 12: Simulated AE events transmitted across the aluminum plate. (a,d) Longitudinally-excited AE sensed by the FBG on the plate, time-domain and frequency-domain results; (b,e) longitudinally-excited AE sensed by the FBG on the ring sensor, time-domain and frequency-domain results; (c,f) Transverse-excited AE sensed by the FBG on the ring sensor, time-domain and frequency-domain results As can be seen in Figure 12b,c the ring sensor showed the capability to sense Lamb waves with propagation paths longitudinal or transverse to the axis of the fiber, showing that the omnidirectionality of the ring sensor applies to AE events. Comparing Figure 12a,c the response of the FBG on the ring sensor was significantly greater than the response of the FBG on the plate (this was true even when comparing strain values by taking the FBG falling slope into account). When examining the FFT of the signals in Figure 12d,e,f, the frequency contents amplified were the first and second resonances of the ring sensor at 82 khz and 265 khz. The larger degree of amplification coming from the second resonance frequency of the ring was surprising, and needs further analysis. Additionally, a low frequency component is clearly visible, interpreted as a flexural-type motion induced in the plate by the application and release of pressure from the pencil. However, upon examination of the time-domain signal, the initial sharp high-frequency peaks were still of a larger magnitude for the FBG on the ring sensor than for the FBG on the plate. From this result, we can conclude that the resonance amplification of the ring was sufficient to amplify an AE signal. 6 CONCLUSIONS This paper has discussed a novel sensor for detection of ultrasonic waves for SHM applications. FEM modeling through harmonic and modal analysis was presented, and sensor placement and ring sensor bonding was addressed in terms of frequency spectrum and performance. A series of pitch-catch, resonance, and acoustic emission experiments were performed across an aluminum plate. From experimental analysis of the free ring sensor, symmetric placement of FBG and PWAS onto the ring sensor was an important factor in its performance. Asymmetrically placed sensors detected extraneous modes near its dominant resonance frequency; this was validated by FEM modeling, which shows additional resonance modes which have nodes

10 along certain lines of symmetry on the ring sensor. A drop in resonance frequency from bonding the ring sensor to a plate was discussed, and interpreted as an effect of having a sensor with a similar stiffness as the substrate it was bonded to. Even so, the ring sensor was still able to produce smooth tuning curves around its dominant resonance. The most important feature of this study was the resonance amplification. This is the first time that this sensor has been shown to be capable of amplifying an incoming Lamb wave. Although the amplification was relatively small, more importantly it validates that the design approach for generating sensors capable of mechanically amplifying a Lamb wave. This is a positive indicator for redesigns which are currently underway, as the inclusion of sensor amplitude as a design factor may have the potential to magnify this effect. Interestingly, the resonance amplification of AE signals was highly significant upon examination of their frequency spectrum. The amplification of the mode near 100 khz was greater than seen in the resonance experiments. Additionally, the second harmonic had a very large amplification. Although, in principle it is desirable to eliminate the effect of the higher harmonics, the larger amplification will be studied as well for redesign and optimization. 7 ACKNOWLEDGEMENTS This material is based on work supported by Office of Naval Research grant # N , Dr Ignacio Perez technical representative. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the Office of Naval Research. 8 REFERENCES [1] Giurgiutiu V., [Structural Health Monitoring with Piezoelectric Wafer Active Sensors, 2 nd Ed.] Elsevier, New York (2014). [2] Giurgiutiu V. and Cuc A., Embedded non-destructive evaluation for structural health monitoring, damage detection, and failure prevention, Shock Vib. Dig. 37(2) (2005). [3] Peters K., Fiber Bragg grating sensors, [Encyclopedia of Structural Health Monitoring], Wiley, New York (2008). [4] Perez I., Cui H., and Udd E., Acoustic emission detection using fiber Bragg gratings, Proc. SPIE 4328, (2001). [5] Giurgiutiu V,. Roman C., Lin B., Frankforter E.. "Omnidirectional piezo-optical ring sensor for enhanced guided wave structural health monitoring," Smart Mater. Struct. 24(1), (2015). [6] Frankforter E., Lin B., and Giurgiutiu V., Piezo-optical measurements for guided wave and acoustic emission structural health monitoring, Proc. SPIE 9062 (2014). [7] Giurgiutiu V., Gresil M., and Roman C. Acousto-ultrasonic sensor Patent Application Publication No. US A1 (2013). [8] Lin B. and Giurgiutiu V., Development of optical equipment for ultrasonic guided wave structural health monitoring, Proc. SPIE 9062 (2014). [9] Norman P. and Davis C., An intensity-based demodulation approach for the measurement of strains induced by structural vibrations using Bragg gratings, Australia: DSTO Defense Science and Technology Organization, DSTO TR (2010).